A Latvian start-up Fintelligence works Like Sherlock Holmes

Fintelligence, a fintech startup, uses artificial intelligence technologies to test the reputation of potential customers and identify the customer‘s level of risk.

Financial institutions are required to verify publicly available information as part of the process of researching potential customers. This process is still a human-intensive one in many financial institutions. To avoid having to do this manually by searching for information on Google, Yandex, and elsewhere, the startup Fintelligence automates this process and displays only the information that may have an impact on the customer’s risk level.

“We analyse all the information we find about a potential customer, identifying the negative information: on scandals, labour law violations, involvement in schemes, tax evasion, etc. We are like the Interpol or the detective Sherlock Holmes for financial institutions, looking for the bad news for them”, says Līga Vinķele, founder of Fintelligence.

Vinķele has worked in banking for more than 12 years. While on parental leave in 2019, she came up with the idea for a customer intelligence system that analyses the information available about customers. After validating the idea with several of Latvia’s largest banks, it was concluded that this is a pressing issue. Especially in different national languages. The whole year 2020 was dedicated to training the AI-based system.

“Training any system of this kind requires a large amount of data, in this case text,” explains Vinķele.

Fintelligence’s target customers are not only banks, but any financial institution. The company’s solution can also be used in procurement processes where there are several bidders and there is a relatively high risk if the selected bidder has a bad reputation or is involved in a financial scheme.

Fintelligence’s solution saves significant financial and time resources for the institutions. Staff do not have to find and read all the information available on internet search engines, but only that which may have a negative impact on the customer’s risk assessment. In the case of opening a business account, it can take up to several weeks to research a client as part of Know Your Customer. For example, if there are 35 pages on Google about a particular company, it takes quite a long time to check them all.

“We can reduce this time by up to 90% by offering to look only at the specific posts that may negatively affect the customer’s rating,” says Vinķele.

Typically, bank staff search for customer information using specific keywords. In this case, the information that is missed may be important. 

“We use natural language processing, distributive semantics and deep linguistic analysis to analyse texts,” she explains.

Fintelligence is currently operating in the X-Europe accelerator programme. This is a European Union-funded programme that focuses more on mentoring participants. Each team has a mentor who helps identify problems and find a product-market fit. The accelerator also introduces different companies that could become partners. 

“With this idea, I was also enrolled in the LIAA-supported mini MBA course at the Riga Business School, which gave me a kick-start in business development,” says Vinķele.

At the beginning of this year, the startup was also selected for the Prototron acceleration programme. 

“We are very pleased to have been selected as one of the top 25 ideas out of a total of 500 applications,” says Vinķele.

So far, the company has developed using its own resources. Fintelligence is currently in discussions with potential investors and plans to attract investment in the near future. 

“We are also actively exploring cooperation opportunities with companies and financial institutions for pilot tests,” says Vinķele.

Source: labsoflatvia.com